In [26]:
#to import data
import pandas as pd
import numpy as np
from sklearn.preprocessing import LabelEncoder
import matplotlib.pyplot as plt
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report, confusion_matrix, accuracy_score
In [2]:
df = pd.read_csv('risk_data.csv')
df.head(10)
Out[2]:
PROSPECTID time_since_recent_payment num_times_delinquent num_times_30p_dpd num_std recent_level_of_deliq PL_enq time_since_recent_enq MARITALSTATUS EDUCATION ... pct_of_active_TLs_ever pct_currentBal_all_TL CC_Flag PL_Flag HL_Flag GL_Flag last_prod_enq2 first_prod_enq2 Credit_Score Approved_Flag
0 1 549 11 0 21 29 6 566 Married 12TH ... 0.200 0.798 0 1 1 0 PL PL 696 P2
1 2 47 0 0 0 0 0 209 Single GRADUATE ... 1.000 0.370 0 0 0 0 ConsumerLoan ConsumerLoan 685 P2
2 3 302 9 0 10 25 0 587 Married SSC ... 1.000 0.585 0 0 1 0 ConsumerLoan others 693 P2
3 4 -99999 0 0 5 0 -99999 -99999 Married SSC ... 1.000 0.990 0 0 0 0 others others 673 P2
4 5 583 0 0 53 0 0 3951 Married POST-GRADUATE ... 0.333 0.000 0 0 0 0 AL AL 753 P1
5 6 245 14 13 5 26 7 7 Married 12TH ... 0.167 0.000 0 0 1 0 ConsumerLoan PL 668 P3
6 7 49 0 0 25 0 -99999 -99999 Married SSC ... 0.667 0.998 0 0 1 0 others others 703 P1
7 8 74 3 2 27 133 0 2 Married UNDER GRADUATE ... 0.333 0.300 0 0 0 0 ConsumerLoan others 676 P2
8 9 424 3 3 0 38 0 108 Married SSC ... 1.000 0.465 0 0 0 0 ConsumerLoan others 658 P4
9 10 39 0 0 39 0 -99999 -99999 Married 12TH ... 0.500 1.121 0 0 0 0 others others 705 P1

10 rows × 24 columns

In [3]:
df.drop(columns=['PROSPECTID','num_times_30p_dpd','last_prod_enq2','first_prod_enq2'], inplace=True)
In [4]:
df.replace(-99999, 0, inplace=True)
df.head()
Out[4]:
time_since_recent_payment num_times_delinquent num_std recent_level_of_deliq PL_enq time_since_recent_enq MARITALSTATUS EDUCATION AGE GENDER NETMONTHLYINCOME Time_With_Curr_Empr pct_of_active_TLs_ever pct_currentBal_all_TL CC_Flag PL_Flag HL_Flag GL_Flag Credit_Score Approved_Flag
0 549 11 21 29 6 566 Married 12TH 48 M 51000 114 0.200 0.798 0 1 1 0 696 P2
1 47 0 0 0 0 209 Single GRADUATE 23 F 19000 50 1.000 0.370 0 0 0 0 685 P2
2 302 9 10 25 0 587 Married SSC 40 M 18 191 1.000 0.585 0 0 1 0 693 P2
3 0 0 5 0 0 0 Married SSC 34 M 10000 246 1.000 0.990 0 0 0 0 673 P2
4 583 0 53 0 0 3951 Married POST-GRADUATE 48 M 15000 75 0.333 0.000 0 0 0 0 753 P1
In [5]:
columns = ['MARITALSTATUS', 'GENDER', 'Approved_Flag', 'EDUCATION']

for column in columns:
    unique_values = df[column].unique()
    print(f"{column} = {' '.join([f'{value}' for value in unique_values])}")

MARITALSTATUS = Married Single
GENDER = M F
Approved_Flag = P2 P1 P3 P4
EDUCATION = 12TH GRADUATE SSC POST-GRADUATE UNDER GRADUATE OTHERS PROFESSIONAL
In [34]:
columns_to_encode = ['MARITALSTATUS', 'GENDER', 'Approved_Flag', 'EDUCATION']

encoder = LabelEncoder()

for column in columns_to_encode:
    df[column] = encoder.fit_transform(df[column])
    
df.head(15).round()
Out[34]:
time_since_recent_payment num_times_delinquent num_std recent_level_of_deliq PL_enq time_since_recent_enq MARITALSTATUS EDUCATION AGE GENDER NETMONTHLYINCOME Time_With_Curr_Empr pct_of_active_TLs_ever pct_currentBal_all_TL CC_Flag PL_Flag HL_Flag GL_Flag Credit_Score Approved_Flag
0 549.0 11 21 29 6 566.0 0 0 48 1 51000.0 114 0.0 1.0 0 1 1 0 696 1
1 47.0 0 0 0 0 209.0 1 1 23 0 19000.0 50 1.0 0.0 0 0 0 0 685 1
2 302.0 9 10 25 0 587.0 0 5 40 1 26519.0 191 1.0 1.0 0 0 1 0 693 1
3 249.0 0 5 0 0 263.0 0 5 34 1 10000.0 246 1.0 1.0 0 0 0 0 673 1
4 583.0 0 53 0 0 3951.0 0 3 48 1 15000.0 75 0.0 0.0 0 0 0 0 753 0
5 245.0 14 5 26 7 7.0 0 0 35 1 26519.0 154 0.0 0.0 0 0 1 0 668 2
6 49.0 0 25 0 0 263.0 0 5 35 1 30000.0 81 1.0 1.0 0 0 1 0 703 0
7 74.0 3 27 133 0 2.0 0 6 30 1 27000.0 126 0.0 0.0 0 0 0 0 676 1
8 424.0 3 0 38 0 108.0 0 5 39 0 18000.0 122 1.0 0.0 0 0 0 0 658 3
9 39.0 0 39 0 0 263.0 0 0 28 1 18000.0 68 0.0 1.0 0 0 0 0 705 0
10 35.0 0 0 0 4 87.0 1 1 27 1 20000.0 95 1.0 1.0 1 1 0 0 685 1
11 2543.0 0 0 0 0 263.0 0 5 57 0 30000.0 141 0.0 0.0 0 0 0 0 682 1
12 1037.0 1 33 31 0 3.0 0 5 30 1 55000.0 126 0.0 0.0 0 0 0 0 669 1
13 46.0 0 0 0 0 164.0 0 5 39 1 25000.0 125 0.0 0.0 0 0 0 0 687 1
14 395.0 0 0 0 0 1135.0 0 0 27 1 30000.0 154 0.0 0.0 0 0 0 0 692 1
In [7]:
df.loc[df['NETMONTHLYINCOME'] < 100, 'NETMONTHLYINCOME'] = 0

print(df['NETMONTHLYINCOME'])

0        51000
1        19000
2            0
3        10000
4        15000
         ...  
51331    18500
51332    25000
51333    18000
51334    12802
51335    16000
Name: NETMONTHLYINCOME, Length: 51336, dtype: int64
In [8]:
columns_to_modify = ['time_since_recent_payment', 'time_since_recent_enq','NETMONTHLYINCOME']

df[columns_to_modify] = df[columns_to_modify].replace(0, pd.NA)
mean_values = df[columns_to_modify].mean()
df.fillna(mean_values, inplace=True)
df.head(10).round()
Out[8]:
time_since_recent_payment num_times_delinquent num_std recent_level_of_deliq PL_enq time_since_recent_enq MARITALSTATUS EDUCATION AGE GENDER NETMONTHLYINCOME Time_With_Curr_Empr pct_of_active_TLs_ever pct_currentBal_all_TL CC_Flag PL_Flag HL_Flag GL_Flag Credit_Score Approved_Flag
0 549.0 11 21 29 6 566.0 0 0 48 1 51000.0 114 0.0 1.0 0 1 1 0 696 1
1 47.0 0 0 0 0 209.0 1 1 23 0 19000.0 50 1.0 0.0 0 0 0 0 685 1
2 302.0 9 10 25 0 587.0 0 5 40 1 26519.0 191 1.0 1.0 0 0 1 0 693 1
3 249.0 0 5 0 0 263.0 0 5 34 1 10000.0 246 1.0 1.0 0 0 0 0 673 1
4 583.0 0 53 0 0 3951.0 0 3 48 1 15000.0 75 0.0 0.0 0 0 0 0 753 0
5 245.0 14 5 26 7 7.0 0 0 35 1 26519.0 154 0.0 0.0 0 0 1 0 668 2
6 49.0 0 25 0 0 263.0 0 5 35 1 30000.0 81 1.0 1.0 0 0 1 0 703 0
7 74.0 3 27 133 0 2.0 0 6 30 1 27000.0 126 0.0 0.0 0 0 0 0 676 1
8 424.0 3 0 38 0 108.0 0 5 39 0 18000.0 122 1.0 0.0 0 0 0 0 658 3
9 39.0 0 39 0 0 263.0 0 0 28 1 18000.0 68 0.0 1.0 0 0 0 0 705 0
In [9]:
df.isnull().sum()
Out[9]:
time_since_recent_payment    0
num_times_delinquent         0
num_std                      0
recent_level_of_deliq        0
PL_enq                       0
time_since_recent_enq        0
MARITALSTATUS                0
EDUCATION                    0
AGE                          0
GENDER                       0
NETMONTHLYINCOME             0
Time_With_Curr_Empr          0
pct_of_active_TLs_ever       0
pct_currentBal_all_TL        0
CC_Flag                      0
PL_Flag                      0
HL_Flag                      0
GL_Flag                      0
Credit_Score                 0
Approved_Flag                0
dtype: int64
In [10]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 51336 entries, 0 to 51335
Data columns (total 20 columns):
 #   Column                     Non-Null Count  Dtype  
---  ------                     --------------  -----  
 0   time_since_recent_payment  51336 non-null  float64
 1   num_times_delinquent       51336 non-null  int64  
 2   num_std                    51336 non-null  int64  
 3   recent_level_of_deliq      51336 non-null  int64  
 4   PL_enq                     51336 non-null  int64  
 5   time_since_recent_enq      51336 non-null  float64
 6   MARITALSTATUS              51336 non-null  int32  
 7   EDUCATION                  51336 non-null  int32  
 8   AGE                        51336 non-null  int64  
 9   GENDER                     51336 non-null  int32  
 10  NETMONTHLYINCOME           51336 non-null  float64
 11  Time_With_Curr_Empr        51336 non-null  int64  
 12  pct_of_active_TLs_ever     51336 non-null  float64
 13  pct_currentBal_all_TL      51336 non-null  float64
 14  CC_Flag                    51336 non-null  int64  
 15  PL_Flag                    51336 non-null  int64  
 16  HL_Flag                    51336 non-null  int64  
 17  GL_Flag                    51336 non-null  int64  
 18  Credit_Score               51336 non-null  int64  
 19  Approved_Flag              51336 non-null  int32  
dtypes: float64(5), int32(4), int64(11)
memory usage: 7.1 MB
In [11]:
df['Credit_Score'].hist(bins=30)
plt.title('Histogram of Credit Score')
plt.xlabel('Credit Score')
plt.ylabel('Frequency')
plt.show()

# Histogram of Age
df['AGE'].hist(bins=30)
plt.title('Histogram of Age')
plt.xlabel('Age')
plt.ylabel('Frequency')
plt.show()
In [12]:
features = ['time_since_recent_payment', 'num_times_delinquent', 'num_std', 'recent_level_of_deliq', 
            'time_since_recent_enq', 'AGE', 'NETMONTHLYINCOME', 'Time_With_Curr_Empr', 'Credit_Score']
correlation_matrix = df[features].corr()

plt.figure(figsize=(10, 8))
sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm', fmt=".2f", square=True)
plt.title('Correlation Heatmap of Features')
plt.show()
In [13]:
for i in range(len(features)):
    for j in range(i + 1, len(features)):
        plt.figure(figsize=(8, 6))
        plt.scatter(df[features[i]], df[features[j]], alpha=0.5)
        plt.title(f'Scatter Plot: {features[i]} vs {features[j]}')
        plt.xlabel(features[i])
        plt.ylabel(features[j])
        plt.grid(True)
        plt.show()
In [19]:
# Define features (X) and target variable (y)
X = df.drop('Approved_Flag', axis=1)
y = df['Approved_Flag']

# Split the data into training and testing sets (80% train, 20% test)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
In [20]:
# Initialize the Random Forest classifier
rf_classifier = RandomForestClassifier(n_estimators=100, random_state=42)

# Train the classifier
rf_classifier.fit(X_train, y_train)
Out[20]:
RandomForestClassifier(random_state=42)
In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
RandomForestClassifier(random_state=42)
In [23]:
# Make predictions on the test data
y_pred = rf_classifier.predict(X_test)

# Evaluate the model
print("Accuracy:", accuracy_score(y_test, y_pred))
print("Confusion Matrix:\n", confusion_matrix(y_test, y_pred))
print("Classification Report:\n", classification_report(y_test, y_pred))

Accuracy: 0.9928905336969225
Confusion Matrix:
 [[1100    0   35    0]
 [   0 6379    0    0]
 [  38    0 1498    0]
 [   0    0    0 1218]]
Classification Report:
               precision    recall  f1-score   support

           0       0.97      0.97      0.97      1135
           1       1.00      1.00      1.00      6379
           2       0.98      0.98      0.98      1536
           3       1.00      1.00      1.00      1218

    accuracy                           0.99     10268
   macro avg       0.99      0.99      0.99     10268
weighted avg       0.99      0.99      0.99     10268
In [24]:
X_test.head()
Out[24]:
time_since_recent_payment num_times_delinquent num_std recent_level_of_deliq PL_enq time_since_recent_enq MARITALSTATUS EDUCATION AGE GENDER NETMONTHLYINCOME Time_With_Curr_Empr pct_of_active_TLs_ever pct_currentBal_all_TL CC_Flag PL_Flag HL_Flag GL_Flag Credit_Score
8564 138.000000 2 0 26 0 33.000000 1 1 31 1 30000.0 124 0.667 0.039 0 0 0 0 654
26313 249.463599 0 0 0 0 262.542052 0 3 26 1 32000.0 125 1.000 1.000 0 0 0 0 671
12212 178.000000 0 0 0 0 1118.000000 1 1 31 1 15000.0 130 0.250 0.000 0 0 0 0 683
37578 68.000000 0 0 0 0 464.000000 0 5 40 0 24000.0 54 1.000 0.305 0 0 0 0 697
21647 1109.000000 0 3 0 1 115.000000 0 1 53 1 20000.0 63 0.000 0.000 0 0 1 0 675
In [27]:
# Get feature importances
feature_importances = rf_classifier.feature_importances_

# Sort feature importances in descending order
indices = np.argsort(feature_importances)[::-1]

# Print the feature ranking
print("Feature ranking:")
for f in range(X.shape[1]):
    print("%d. feature %d (%f)" % (f + 1, indices[f], feature_importances[indices[f]]))

Feature ranking:
1. feature 18 (0.769125)
2. feature 5 (0.051822)
3. feature 2 (0.046534)
4. feature 4 (0.018559)
5. feature 3 (0.014671)
6. feature 12 (0.014571)
7. feature 8 (0.014168)
8. feature 0 (0.012152)
9. feature 1 (0.011246)
10. feature 11 (0.010996)
11. feature 13 (0.010705)
12. feature 10 (0.009423)
13. feature 7 (0.004270)
14. feature 16 (0.002699)
15. feature 17 (0.002572)
16. feature 15 (0.002203)
17. feature 6 (0.001818)
18. feature 14 (0.001287)
19. feature 9 (0.001177)
In [30]:
df.head()
Out[30]:
time_since_recent_payment num_times_delinquent num_std recent_level_of_deliq PL_enq time_since_recent_enq MARITALSTATUS EDUCATION AGE GENDER NETMONTHLYINCOME Time_With_Curr_Empr pct_of_active_TLs_ever pct_currentBal_all_TL CC_Flag PL_Flag HL_Flag GL_Flag Credit_Score Approved_Flag
0 549.000000 11 21 29 6 566.000000 0 0 48 1 51000.000000 114 0.200 0.798 0 1 1 0 696 1
1 47.000000 0 0 0 0 209.000000 1 1 23 0 19000.000000 50 1.000 0.370 0 0 0 0 685 1
2 302.000000 9 10 25 0 587.000000 0 5 40 1 26518.684515 191 1.000 0.585 0 0 1 0 693 1
3 249.463599 0 5 0 0 262.542052 0 5 34 1 10000.000000 246 1.000 0.990 0 0 0 0 673 1
4 583.000000 0 53 0 0 3951.000000 0 3 48 1 15000.000000 75 0.333 0.000 0 0 0 0 753 0
In [43]:
new_inquary_result = {
    'time_since_recent_payment': 520,
    'num_times_delinquent': 3,
    'num_std': 0,
    'recent_level_of_deliq': 21,
    'PL_enq': 0,
    'time_since_recent_enq': 70,
    'MARITALSTATUS':0,
    'EDUCATION': 1,
    'AGE': 30,
    'GENDER': 0,
    'NETMONTHLYINCOME': 17000,
    'Time_With_Curr_Empr': 115,
    'pct_of_active_TLs_ever': 1,
    'pct_currentBal_all_TL': 0.465,
    'CC_Flag': 0,
    'PL_Flag': 1,
    'HL_Flag': 1,
    'GL_Flag': 0,
    'Credit_Score': 640
}

# Convert individual feature data into DataFrame
individual_df = pd.DataFrame([new_inquary_result])

# Preprocess the individual feature data (e.g., encoding categorical variables)
# Use the same preprocessing steps applied to the training data

# Make predictions using the trained model
prediction = rf_classifier.predict(individual_df)

# Display the prediction
print("Prediction:", prediction)

Prediction: [3]